Three weeks after OpenAI's pricey GPT-6 Astra, Anthropic answered with the opposite pitch: Claude Opus 5.5 delivers frontier performance for less money, not more. Released September 22, it's the first model in Anthropic's new 5.5 family, and it lands squarely on the value-per-capability line rather than chasing the top of the price chart.
Anthropic says Opus 5.5 performs at the level of its Fable 5.1 flagship on most work, while costing about 40% less to run than Opus 5 on typical workloads and generating output more than 30% faster. Pricing is $4 per million input tokens and $20 per million output, 20% below Opus 5, with cache reads down to $0.20, a 60% cut.
The agentic headline
The number that matters most is on agents. On Terminal-Bench 4.0, a test of long, multi-step coding-in-a-terminal work, Opus 5.5 scored 66.4% at high effort, ahead of the 57.9% Anthropic cites for OpenAI's GPT-6 Astra. For anyone running coding agents (in Claude Code or elsewhere), that combination of higher agentic accuracy and lower cost is the whole ballgame.
What it means
The two big labs are now pulling in different directions: OpenAI is selling the most capable model at a premium, Anthropic is selling near-equal capability at a discount. For most real work, especially agents, where you burn a lot of tokens, cheaper-and-faster wins, which is exactly the lane Opus 5.5 is built for. It's available on AWS, Google Cloud, Azure, and the Claude Platform. See where it sits in our best AI models ranking.